The shortlist is now being written before your brand is even in the room, as buyers turn to AI to assist with their research. An advisor asks ChatGPT to compare custody providers. An analyst builds a vendor list in Perplexity. A procurement tool scores RFP responses before a person reads them. In each case, an AI agent decides which brands make the shortlist and how they're described, long before anyone fills out your contact form.

That's why every marketer now needs to treat AI agents as a core audience, sitting right alongside the buyer personas you already know. 

This is the first of two pieces on what that means. 

Part 1 covers marketing to AI: how to make sure your brand shows up, is understood, and gets chosen when agents are doing the research. 

Part 2 covers marketing with AI: how to give the AI tools your own team uses a reliable source of truth.

How buying has changed

AI has moved into the middle of the purchase process faster than most marketing plans have adjusted. Forrester's 2026 State of Business Buying found that 94% of nearly 18,000 business buyers use AI during their buying process, and 6sense's 2025 Buyer Experience Report landed on the same 94% figure. 6sense also found that buyers choose a vendor from their Day One shortlist 95% of the time, and that list forms before any vendor contact. Looking ahead, Gartner predicts that by 2028, 90% of B2B buying will be AI agent intermediated.

Put those together and the stakes are clear. If your brand isn't present and accurately represented in AI answers, you can be eliminated before a human ever engages. And most of that filtering happens on surfaces you don't control: earned media, review sites, analyst coverage, and forums like Reddit.

The insight: agents are an audience, not just a channel

It's tempting to file all of this under SEO and hand it to the web team. We'd argue it belongs higher up. The agent is a reader with its own needs and habits, so it deserves the same treatment you give a CFO or a head of marketing: a persona in your customer analysis, questions in your stakeholder interviews, and a seat in your messaging and content strategy.

Here's what that persona looks like in practice:

• It rewards verifiable facts. In the KDD 2024 GEO study, tactics like adding statistics, quotations, and citations lifted visibility in AI answers by up to 40%, and citing sources raised visibility for fifth-ranked pages by 115%.
• It trusts other people's words more than yours. Muck Rack's May 2026 research, based on more than 25 million links, found earned media drives 84% of AI citations, while paid and advertorial content accounts for just 0.3%.
• It favors brands it already knows. In a study of 670 trials with otherwise identical products, LLMs recommended the recognized brand every time. Ads won't put you in an AI answer directly, but they do build awareness with the people who write the reviews, Reddit threads, and articles that agents read.
• It reads literally. Your homepage headline, your boilerplate, and your About page are all taken at face value. "We empower teams to dream bigger" tells an agent nothing. "B2B custody platform for RIAs managing $100M+ AUM" tells it everything it needs.
• It has no patience limit. It will read your IR portal, page ten of search results, and a three-year-old Reddit thread about your fees that no human buyer would open. Anything public about your brand is source material.

The catch: people still make the final call

None of this means writing for robots. Agents build the shortlist, but people still make the final decision, and emotion, aesthetics, and craft build the recognition and preference that agents end up rewarding. The goal is dual optimization: work that resonates with people and holds up under a machine's literal reading. Great marketing is still first-principles marketing. What's new is a second reader who takes everything at face value.

What to do about it

The principles that have always driven great marketing still apply. You need to be found, be understood, and be preferred. Here's how each one plays out when an agent is part of your audience.

Be Found

1. Make your site easy for machines to read

An agent can't recommend a brand it can't locate or parse. Start with clean site code, schema markup, crawlable pages, a /for-AI page, and clear policies for AI crawlers. Then build content for how people now ask questions: FAQ-style pages, the answer in the first 30% of the page, and the comparison content agents go looking for.

2. Show up where agents do their reading 

Most discovery happens off your website. Map the publications, analysts, review platforms, and forums that AI models actually cite in your category, then point your digital PR at them. Earned coverage has become retrieval data for the models, which makes PR one of the highest-leverage AI visibility channels you have.

Be Understood

3. Say plainly what you do

Write your positioning in two registers: a conceptual line for people and a plain, definitional version for machines. On owned channels, pair conceptual headlines with plain-spoken subheads, make your H1s say what the company does, and label navigation by what's behind it. Save the bigger creative swings for advertising and campaigns, where they do their best work with people.

4. Fill the gaps before someone else does

When information about your brand is missing, AI fills the gap with third-party guesses. This matters most in regulated categories, where firms often can't publish fees or performance. Counter it with self-contained, dated, disclaimer-bearing statements that hold up when an engine lifts a single sentence, plus entity data that states key facts (like fiduciary status) clearly. It's also worth auditing what's already public about you, since outdated pages and old threads are all fair game for an agent.

Be Preferred

5. Give agents a reason to prefer you

Getting into a model's preference set, the short list of brands it recommends, comes down to proof. Publish original research and proprietary data, back specific claims with evidence, and build consistent third-party corroboration across the web. Keep investing in brand awareness, too. The better known you are, the more latitude agents give you.

Measure what you can

There's no Search Console for prompts yet, and AI answers vary from one run to the next. Run a fixed set of prompts repeatedly against a frozen competitor list to track your AI share of voice and citation rate. Pair that with AI referral traffic, branded search velocity, and a simple "Did you use AI in your research?" question on your intake forms.

The bottom line

Buyers haven't stopped needing great brands. They've added a research assistant that reads everything, takes it literally, and trusts what others say about you. Treat that assistant as an audience, and your brand earns its place on the shortlist. Then the work that moves people can do what it's always done.

In Part 2, we'll turn to the AI already inside your organization, and how to make sure it represents your brand as well as your best people do.

Want to know how your brand shows up when an AI agent is doing the research?

‍